CN103476080A - Self-adaption vertical handoff method based on residence time - Google Patents

Abstract

The invention provides a self-adaption vertical handoff method (MAAV) based on residence time, wherein the residence time, within the coverage range of a certain base station, of a user is taken into consideration on the basis of a vertical handoff method based on multi-attribute decision making. According to the self-adaption vertical handoff method based on the residence time, user motion information is used for estimating the residence time, within the coverage range of the certain base station, of the user, and weight corresponding to the residence time is adjusted dynamically according to the residence time, so that when the residence time meets handoff conditions, influences from the residence time become less as the residence time increases, and at last, a simple weighing method (SAW) is used for combining the residence time with the existing multi-attribute vertical handoff method for conducting judgment. Compared with the existing vertical handoff method based on multi-attribute decision making, the self-adaption vertical handoff method can effectively reduce unnecessary handoff times, and improve the service quality.

Description

Adaptive vertical handoff method based on the residence time

Technical field:

The invention belongs to the communications field, be specifically related to heterogeneous network switching field.

Background technology

Next generation network is the network of integrated current various wireless access technologys, and most important feature is exactly isomerism, can show as the difference of various wireless access technologys on access bandwidth, Time Delay of Systems, expenses standard, security performance etc.Moreover, user's business demand is also that diversified, different business is also different to the requirement of systematic function.Therefore, different from horizontal handover decisions, the factor that the decision-making of vertical switching need to be considered is a lot, comprises type of service, service rate, network state, systematic function, fail safe, translational speed and user preference etc., belongs to uncertain multiattribute decision problem.Switch determining method based on signal strength signal intensity RSS in horizontal switchover policy be not applicable to vertical switching.The effective utilization that needs the efficient feasible switch determining method of research and practical control mode to reach systematic function optimization and resource.

Due to heterogeneous network need to consider a lot, thereby the switch decision problem that solves heterogeneous network is finally a multiattribute decision problem.Such as based on receiving signal strong and weak (RSS), network charges etc., selecting network [the LING Yu-tao that will switch, YI Ben-shun, ZHU Qiu-ping.Vertical handoff decision strategy in wireless overlay networks[C]. //Proceedings of the 5th International Conference on Wireless Communications, Networking and Mobile Computing.2009:1-3]; Select network [Majed Haddad by user utility function and network system optimal policy, Salah Eddine Elayoubi, Eitan Altman.A.Hybrid Approach for Radio Resource Management in Heterogeneous Cognitive Networks[J] .IEEE JOURNAL ON SELECTED AREAS IN COMMUNICATIONS, VOL.29, NO.4, APRIL2011]; Adopt the dynamic weighting system of selection to obtain cost function [the Nie J of interface, He x, Zhou z, et al.Benefit-driven handoff between WMAN and WLAN[C] //Proceedings of the2005IEEE Military Communications Conference (MILCOM2005), Atlantic City, NJ, USA, 2005:2223-2229].These are all the concrete research to the multiattribute judgement, its main thought is to consider the important property of impact judgement for the characteristic of network, set different weighted values according to the characteristics of property parameters, the benefit/cost function of computing network is as the foundation of vertical switch decision.

Existing multiattribute decision method is not considered user's motion state, and user's motion state is reflected in the user on the residence time in base station range, and this can impact switch decision.As the back and forth movement of user at the edge, base station, along the circular motion at edge, base station etc., if do not consider in this case, forms of motion can cause unnecessary switching, reduces service quality.The user movement correlative factor is taken into account, and combined and carry out switch decision and can reduce switching times with the network performance correlative factor, improve service quality.

Document [Nancy S, Ahmed K.A mobility prediction architecture based on contextual knowledge and spatial conceptual maps[J] .IEEE Transactions on Mobile Computing, 2005,4 (6): 537-551] adopted the method based on cost function in.Scheme has been considered the performance parameter of heterogeneous networks, as access bandwidth, time delay, expense, fail safe, can give full play to the advantage of various access networks, realizes preferably best link.The method has been carried out normalized to parameters, utilizes simple weighted method SAW summation, and relatively the cost of heterogeneous networks, select optimal switching objective network.With the vertical handoff method based on RSS, compare, the method has taken into full account the difference of network performance, has increased throughput of system, has reduced service fee, has improved the utilance of system resource.

Although the performance parameter that the method adopts the multiattribute judgement to consider heterogeneous networks, as signal to noise ratio snr, cost, bandwidth etc., has improved throughput of system and resource utilization.But, user's moving situation is not taken into account, under some motion conditions of user, " table tennis " switching just may occur in (as back and forth movement), and this thing happens can allow the performance degradation of system for a large number of users, and the service quality received also can reduce.

Document [Feng He, Furong Wang.Position aware vertical handoff decision algorithm in heterogeneous wireless networks[C] .International Conference on Wireless Communications, Networking and Mobile Computing, 2008:1-5] adopt the vertical handoff method of motion prediction.The method emphasis has been considered the impact that movement velocity, direction, the motion model of terminal cause switching.Estimate the variable in distance between terminal and base station according to the variation of the intensity RSS that receives signal.History information estimation movement velocity by RSS, then estimate the residence time of terminal in certain base station range, and this can reduce unnecessary switching, improves communication quality.

The characteristic of having ignored system although the method is considered user's motion state.Widely different due to network performance in heterogeneous network, the differences such as its bandwidth provided, service quality, rate, so the method can not allow the performance of system reach optimization, the resource utilization of system can not reasonably be utilized.The method is only applicable to simple heterogeneous network situation, for complicated heterogeneous network handoff environment, can not guarantee that the user can access best network in real time, can not reasonable distribution heterogeneous network resource, cause like this wasting of resources to a certain degree.

Summary of the invention

Goal of the invention: proposed the vertical handover scheme of self adaptation based on the residence time.Considered two class parameters in this scheme, a class is the parameter (meaning with the residence time) of user movement Determines; Equations of The Second Kind is the parameter (as access bandwidth, time delay, expense etc.) that the wireless access technology relevant to network performance determines.This two classes parameter directly affects the performance of vertical switching.The present invention, in conjunction with two class parameters advantage separately, all takes relevant parameter into account, and dynamically adjusts the weight of different parameters, has proposed new adaptive vertical handoff method.The method can be avoided " table tennis " effect, and when reducing unnecessary switching, performance that can taking into account system, improve user's service quality.

Technical scheme:

At first to carry out the residence time of estimating subscriber's in base station range by user state information, then calculate the weight of residence time parameter.Carry out normalized with other switch decision parameter again, finally obtain the community set for vertical switching, calculate benefit function according to the value of each attribute and judge whether to be switched.

The detailed method of the adaptive vertical handoff method based on the residence time (MAAV) is described below:

The first step, the broadcast singal of the peripheral base station/WAP (wireless access point) detected according to user terminal, if received signal strength RSS is greater than certain threshold value, join this base station candidate network set B S,

BS＝{BS ₁，BS ₂，...，BS _N} (1)

RSS _i＞RSS _i，thres (2)

Wherein, RSS _ifrom base station BS _iacknowledge(ment) signal intensity; RSS _{i, thtes}it is the RSS minimum value that proper communication is carried out in terminal and this base station.Different wireless access technologys, RSS _{i, thres}may be different.

Second step, utilizing the GPS-MAV method is the residence time set T of GPS positional information calculation user in each candidate base station coverage,

$T=\{t_{\mathrm{dwell}}^1,t_{\mathrm{dwell}}^2,...,t_{\mathrm{dwell}}^N\}---\left(3\right)$

In formula,

that the user is at base station BS _iresidence time in coverage.Concrete grammar is as follows:

As Fig. 5, the position of WLAN base station is an O, and its coordinate is (x ₀, y ₀); The current location of terminal MN is a D, and its coordinate is (x _n, y _n); The present speed of terminal is big or small v _t, its direction and vector

identical; The definite ray of base station and terminal is its middle conductor OD is the distance between base station and terminal; Ray

identical with x axle direction; The terminal direction of motion

with x axle forward

angulation is and

base station and terminal line

with x axle forward

angle be θ _t, and 0≤θ _t<2 π;

To terminal velocity v _tdo resolution of vectors; Its the component size of direction is v _bS->MN, the component size of its vertical direction is v _vertical; Velocity component v _verticalnot affecting the distance between base station and terminal, is v apart from being subject to velocity component between base station and terminal _bS->MNimpact; In addition, v' is another speed constantly of terminal, the angle corresponding with v'; According to top hypothesis, can express velocity component v _bS->MNand v _vertical,

If v _bS->MN>0, the distance between terminal and base station increases, and received signal strength RSS reduces gradually; If v _bS->MN<0, the distance between terminal and base station reduces, and received signal strength RSS increases gradually;

Angle theta _tit is vector

with vector

the angle become, be (x _n-x ₀, y _n-y ₀) angle definite with (1,0); So

${\mathrm{\&theta;}}_t=\left\{\begin{array}{c}\arccos \left(\frac{x_N-x_0}{\sqrt{{(x_N-x_0)}^2+{(y_N-y_0)}^2}}\right),(y_N-y_0)\&GreaterEqual;0\\ 2\mathrm{\&pi;}-\arccos \left(\frac{x_t-x_0}{\sqrt{{(x_N-x_0)}^2+{(y_N-y_0)}^2}}\right),(y_N-y_0)<0\end{array}\right.---\left(6\right)$

The end coordinates of previous moment is (x _n-1, y _n-1); Because the time interval that adjacent twice position upgraded is very little, can think that terminal does linear uniform motion within this cycle; Thereby the direction of motion of terminal just can be by the position coordinates (x in these two moment in one-period _n, y _n) and (x _n-1, y _n-1) determine,

Consider the randomness of user movement, carrying out speed while decomposing, employing be angle in nearest a period of time

mean value

In formula, α is index distribution smoothing factor,

by positional information p _kand p _k-1calculate;

The current distance d of terminal and WLAN base station _tfor

$d_t=\sqrt{{(x_N-x_0)}^2+{(y_N-y_0)}^2}---\left(9\right)$

The positional information p that terminal is up-to-date _nand p _n-1movement velocity size v _tfor,

$v_t=\frac{\left|\mathrm{\&Delta;d}\right|}{\mathrm{\&Delta;t}}=\frac{\sqrt{{(x_N-x_{N-1})}^2+{(y_N-y_{N-1})}^2}}{|t_N-t_{N-1}|}---\left(10\right)$

Consider the randomness of user movement, carrying out speed while decomposing, employing be velocity magnitude v in a period of time _tmean value

${\tilde{v}}_t=(1-\mathrm{\&alpha;})\underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}^{N-k}{v}_k---\left(11\right)$

In formula, v _kby positional information p _kand p _k-1calculate;

So speed exists

the component size v of direction _bS->MNcomponent size can be expressed as

If GPS updating location information cycle T _gPSvery little and can obtain continuously sample value, so, angle mean value

velocity magnitude mean value

just can be expressed as,

${\tilde{v}}_t=(1-\mathrm{\&alpha;}){\&Integral;}_0^N{\mathrm{\&alpha;}}^{N-i}{v}_i\mathrm{di}---\left(14\right)$

It is not constant that the randomness of user movement makes velocity magnitude and direction; A kind of situation is exactly to cover edge at certain point base stations to do back and forth movement; Now, the average speed size of terminal

value may be very large, but the displacement within a period of time is very little; If while according to above formula, asking the residence time, the result obtained has very large error, at this, proposes a new argument, i.e. effective displacement factor-beta,

$\mathrm{\&beta;}=\frac{|d_N-{\mathrm{<figure-callout\; id="0"\; label="d"\; filenames="HDA0000379793210000022.png,HDA0000379793210000031.png"\; state="\{\{state\}\}">d</figure-callout>}}_0|}{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{dis}}_{i,i-1}}---\left(15\right)$

${\mathrm{dis}}_{i,i-1}=\sqrt{{(x_i-x_{i-1})}^2+{(y_i-y_{i-1})}^2}---\left(16\right)$

Effectively the displacement factor-beta has reacted the scale that terminal moves with respect to origin-location;

Estimate the residence time t of user in the current base station coverage _dwell,

$t_{\mathrm{dwell}\_\mathrm{dispose}}=\frac{R_{\mathrm{WLAN}}-d_t}{\left|v_{\mathrm{BS}->\mathrm{MN}}\right|}---\left(17\right)$

$t_{\mathrm{dwell}\_\mathrm{factor}}=\frac{R_{\mathrm{WLAN}}-d_t}{\mathrm{\&beta;}\left|{\tilde{v}}_t\right|}---\left(18\right)$

t _dwell＝min{t _{dwell_dispose}，t _{dwell_factor}} (19)

The 3rd step, according to the user at certain candidate base station BS _ithe residence time of coverage

calculate the weight α of using in comprehensive judgement _i, it can be expressed as

${\mathrm{\&alpha;}}_i=\exp (-t_{\mathrm{dwell}}^i/t_{\mathrm{handoff}\_\mathrm{delay}})---\left(20\right)$

Wherein, t _{handoff_delay}it is handover delay.So, candidate base station BS _ithe aggreggate utility value

just can be expressed as,

$U_{\mathrm{total}}^i={\mathrm{\&alpha;}}_i{f}_{\mathrm{dwell}}^i+(1-{\mathrm{\&alpha;}}_i)U_{\mathrm{performance}}^i---\left(21\right)$

$f_{\mathrm{dwell}}^i=\left\{\begin{array}{c}-\frac{1}{{\mathrm{\&alpha;}}_i},t_{\mathrm{dwell}}^i\&le;t_{\mathrm{handoff}\_\mathrm{delay}}\\ 1,t_{\mathrm{dwell}}^i>t_{\mathrm{handoff}\_\mathrm{delay}}\end{array}\right.---\left(22\right)$

Wherein,

bS _ithe performance value of utility,

it is the residence time

the standardization value.Visible, the residence time

larger, α _iless, the residence time pair

impact less, the performance value of utility

right

impact larger.That is to say, the residence time hour,

right play a major role; When larger in the residence time, the performance value of utility right

play a major role.Work as the residence time

be less than handover delay

the time, the aggreggate utility value

be less than zero, i.e. base station BS _ialmost possibility is not chosen as handover-target base station.

The 4th step, the normalized network performance parameter, as bandwidth, time delay, expense, and utilize simple weighted method SAW to calculate the performance value of utility of each candidate base station

$U_{\mathrm{performance}}^i={\mathrm{\&omega;}}_B{f}_B^i+{\mathrm{\&omega;}}_{\mathrm{\&tau;}}{f}_{\mathrm{\&tau;}}^i+{\mathrm{\&omega;}}_C{f}_C^i---\left(23\right)$

Wherein, ω _b, ω _τ, ω _cbe respectively the weight of bandwidth, time delay, expense,

the normalized value corresponding with it.Composite type 5 and formula 6, just can obtain the aggreggate utility value

$U_{\mathrm{total}}^i={\mathrm{\&alpha;}}_i+(1-{\mathrm{\&alpha;}}_i){\mathrm{\&omega;}}_B{f}_B^i+(1-{\mathrm{\&alpha;}}_i){\mathrm{\&omega;}}_{\mathrm{\&tau;}}{f}_{\mathrm{\&tau;}}^i+(1-{\mathrm{\&alpha;}}_i){\mathrm{\&omega;}}_C{f}_C^i---\left(24\right)$

In addition, bandwidth, time delay, the isoparametric unit of expense difference, and also the span of the same parameters of different networks also differs larger, and this has caused very large difficulty just to comparison and overall merit between these different parameters.So this method has been carried out normalized to these parameters, make normalized numerical value be not more than 1.In addition, bandwidth is larger, and service quality QoS is better; Time delay and expense are less, and it is better that the user experiences; Value of utility U _ilarger, the performance of this network is better.Expense is generally fixed, and can be made as constant.

Standardization value for bandwidth

can be expressed as,

$f_B^i=\left\{\begin{array}{c}-\frac{1}{(1-\mathrm{\&alpha;}){\mathrm{\&omega;}}_B},B_i\&le;B_{\min }\\ \frac{B_i-B_{\min }}{B_{\max }-B_{\min }},B_{\min }<B_i<B_{\max }\\ 1,B_i\&GreaterEqual;B_{\max }\end{array}\right.---\left(25\right)$

Standardization value for time delay

can be expressed as,

$f_{\mathrm{\&tau;}}^i=\left\{\begin{array}{c}-\frac{1}{(1-\mathrm{\&alpha;}){\mathrm{\&omega;}}_{\mathrm{\&tau;}}},{\mathrm{\&tau;}}_i\&GreaterEqual;{\mathrm{\&tau;}}_{\max }\\ \frac{{\mathrm{\&tau;}}_{\max }-{\mathrm{\&tau;}}_i}{{\mathrm{\&tau;}}_{\max }-{\mathrm{\&tau;}}_{\min }},{\mathrm{\&tau;}}_{\min }<{\mathrm{\&tau;}}_i<{\mathrm{\&tau;}}_{\max }\\ 1,{\mathrm{\&tau;}}_i\&le;{\mathrm{\&tau;}}_{\min }\end{array}\right.---\left(26\right)$

From formula 7 and formula 8, can find out, the standardization value of bandwidth and time delay all is not more than 1.When bandwidth/time delay does not meet the Minimum requirements of user's application, the standardization value substitution formula 7 by now, just can obtain the aggreggate utility value

minus.I.e. aggreggate utility value now

very little, the possibility that is chosen as handover-target base station is very little.It is very good that this just can be avoided user terminal to be switched to most of performance parameter effectively, and go in the candidate base station that only has a parameter not meet consumers' demand.

The 5th step, utilize step 3 and step 4, obtains respectively each base station BS in candidate network set B S _icorresponding aggreggate utility value

and the aggreggate utility value of Current Serving BTS

if Current Serving BTS is not the optimal service base station, need to carry out handoff procedure.

$U_{\mathrm{total}}^{t\arg \mathrm{et}}=\max \left\{U_{\mathrm{total}}^i\right\}---\left(27\right)$

Wherein, for the aggreggate utility maximum, its corresponding base station is BS _target.MAAV method flow in this paper as shown in Figure 1.

Beneficial effect

(1) this patent has proposed the adaptive vertical handoff method based on the residence time, and this scheme has been considered user's motion state in the multiattribute judgement, and the weights that residence time that characterizes the user movement state calculated are introduced the multiattribute judgement.Simulation result shows can reduce the covering unnecessary switching in edge, improves the service quality of system.

(2) adaptive method for switching that this patent proposes can dynamically be adjusted corresponding weight according to the residence time, and the residence time, large weight was little, other its Main Functions of judgement attribute, and vice versa.Also the user movement state is taken into account when considering overall system performance like this, can be adapted to the variation of handoff environment.

The accompanying drawing explanation

The adaptive vertical handoff method flow chart of Fig. 1 based on the residence time

Fig. 2 user movement trajectory diagram

Switching times under Fig. 3 different distance relatively

QoS under Fig. 4 different distance EM relatively

Fig. 5 user velocity resolution of vectors figure

Simulation analysis

The network model that the artificial network environment adopts is 1 UMTS base station and 1 WLAN WAP (wireless access point).One has 100 users.Each user's movement velocity is 3m/s, from a S, does linear uniform motion to an E, then turns back to a S from an E.Point M is that the WLAN WAP (wireless access point) covers the point on edge, and the length of line segment SM is 20m; The minimum value of the length of line segment EM is 0, and maximum is L _max, obedience is uniformly distributed.The user movement track as shown in Figure 2.

Customer service to the requirement of network performance is: (1) bandwidth is minimum for 1Mbps, is 6Mbps to the maximum; (2) time delay is 50ms to the maximum, and minimum is that 20ms (3) service fee is the smaller the better.The bandwidth of UMTS network is 2Mbps, and time delay is 20ms, and service fee is 0.8; The bandwidth of wlan network is 11Mbps, and time delay is 41ms, and service fee is 0.4.In addition, the weight of bandwidth, time delay, expense is followed successively by ω _b=0.5, ω _τ=0.2, ω _c=0.3.

Below compared the adaptive vertical handoff method based on the residence time (MAAV) and the existing vertical handoff method (SAW) based on the simple weighted method that this patent proposes, and compared on switching times and QoS of customer QoS.Wherein, switching times more as shown in Figure 2.

As can be seen from Figure 3, along with the peaked increase of distance EM, MAAV and the corresponding switching times of SAW method all constantly increase, but the switching times of MAAV method is fewer than the switching times of SAW method all the time.When this is smaller because of the distance that enters the wlan network coverage as the user, does not switch, otherwise will cause " ping-pong ", even can cause conversation.The MAAV method has been considered the user movement factor just, has avoided this situation, thereby reduces unnecessary switching times.

As seen from Figure 4, at ultimate range EM hour, the service quality QoS of MAAV and SAW method constantly descends, and this is because the increase of switching times causes the decline of service quality.Then along with the increase of ultimate range EM, the service quality QoS of MAAV and SAW method constantly rises, and the QoS gap of two kinds of methods is more and more less, this is that to be used in the time be in wlan network longer because make apart from the increase of EM, and the performance advantage of wlan network has compensated the QoS loss that switching causes.But the service quality of MAAV method is higher than the service quality of SAW method all the time.So MAAV method in this paper can take into full account user's requirement, improve QoS of customer.

Complex chart 3 and Fig. 4, the adaptive vertical handoff method of based on motion perception (MAAV) is compared with the existing vertical handoff method (SAW) based on multiple attribute decision making (MADM), not only can reduce unnecessary switching, avoid " ping-pong ", can also improve QoS of customer QoS.In addition, the method is convenient to be combined with existing various multiattribute decision methods, improves the switch decision performance.

Embodiment

As shown in Figure 1, implementation step is as follows for all handoff procedures:

Step 1, the broadcast singal of the peripheral base station/WAP (wireless access point) detected according to user terminal, if received signal strength RSS is greater than certain threshold value, join this base station candidate network set B S={BS ₁, BS ₂..., BS _n.

Step 2, utilize the residence time set of GPS positional information calculation user in each candidate base station coverage

$T=\{t_{\mathrm{dwell}}^1,t_{\mathrm{dwell}}^2,...,t_{\mathrm{dwell}}^N\}.$

Step 3, according to the user at certain candidate base station BS _ithe residence time of coverage

calculate the weight α of using in comprehensive judgement _i.

Step 4, the normalized network performance parameter, as bandwidth, time delay, expense, and utilize simple weighted method SAW to calculate the performance value of utility of each candidate base station.

Step 5, utilize step 3 and step 4, obtains respectively each base station BS in candidate network set B S _icorresponding aggreggate utility value

and the aggreggate utility value of Current Serving BTS

if Current Serving BTS is not the optimal service base station, need to carry out handoff procedure.

Claims (2)

1. the adaptive vertical handoff method based on the residence time, is characterized in that, the residence time by the user in base station range adds in the vertical switch decision of multiattribute, and calculate its weight; Residence time attribute and other judgement attributes are combined, dynamically adjust each attribute weight to adapt to different switching situations; The step of the method is as follows:

Step 1, the broadcast singal of the peripheral base station/WAP (wireless access point) detected according to user terminal, if received signal strength RSS is greater than certain threshold value, join this base station candidate network set B S,

BS={BS ₁，BS ₂，...，BS _N} (1)

RSS _i>RSS _i，thres (2)

Wherein, RSS _ifrom base station BS _iacknowledge(ment) signal intensity; RSS _{i, thres}it is the RSS minimum value that proper communication is carried out in terminal and this base station; Different wireless access technologys, RSS _{i, thres}may be different;

Step 2, utilize the residence time set T of GPS positional information calculation user in each candidate base station coverage,

$T=\{t_{\mathrm{dwell}}^1,t_{\mathrm{dwell}}^2,...,t_{\mathrm{dwell}}^N\}---\left(3\right)$

In formula,

that the user is at base station BS _iresidence time in coverage;

Step 3, according to the user at certain candidate base station BS _ithe residence time of coverage

calculate the weight α of using in comprehensive judgement _i, it can be expressed as

${\mathrm{\&alpha;}}_i=\exp ({-t}_{\mathrm{dwell}}^i/t_{\mathrm{handoff}\_\mathrm{delay}})---\left(4\right)$

Wherein, t _{handoff_delay}it is handover delay; So, candidate base station BS _ithe aggreggate utility value

just can be expressed as,

$U_{\mathrm{total}}^i={\mathrm{\&alpha;}}_i{f}_{\mathrm{dwell}}^i+(1-{\mathrm{\&alpha;}}_i)U_{\mathrm{performance}}^i---\left(5\right)$

$f_{\mathrm{dwell}}^i=\left\{\begin{array}{c}-\frac{1}{{\mathrm{\&alpha;}}_i},t_{\mathrm{dwell}}^i\&le;t_{\mathrm{handoff}\_\mathrm{delay}}\\ 1,t_{\mathrm{dwell}}^i>t_{\mathrm{handoff}\_\mathrm{delay}}\end{array}\right.---\left(6\right)$

Wherein,

bS _ithe performance value of utility,

it is the residence time the standardization value;

Step 4, the normalized network performance parameter, as bandwidth, time delay, expense, and utilize simple weighted method SAW to calculate the performance value of utility of each candidate base station

$U_{\mathrm{performance}}^i={\mathrm{\&omega;}}_B{f}_B^i+{\mathrm{\&omega;}}_{\mathrm{\&tau;}}{f}_{\mathrm{\&tau;}}^i+{\mathrm{\&omega;}}_C{f}_C^i---\left(7\right)$

Wherein, ω _b, ω _τ, ω _cbe respectively the weight of bandwidth, time delay, expense, the normalized value corresponding with it; Composite type 5 and formula 6, just can obtain the aggreggate utility value

$U_{\mathrm{total}}^i={\mathrm{\&alpha;}}_i+(1-{\mathrm{\&alpha;}}_i){\mathrm{\&omega;}}_B{f}_B^i+(1-{\mathrm{\&alpha;}}_i){\mathrm{\&omega;}}_{\mathrm{\&tau;}}{f}_{\mathrm{\&tau;}}^i+(1-{\mathrm{\&alpha;}}_i){\mathrm{\&omega;}}_C{f}_C^i---\left(8\right)$

In addition, bandwidth, time delay, the isoparametric unit of expense difference, and also the span of the same parameters of different networks also differs larger, and this has caused very large difficulty just to comparison and overall merit between these different parameters; So this method has been carried out normalized to these parameters, make normalized numerical value be not more than 1;

Standardization value for bandwidth

can be expressed as,

$f_B^i=\left\{\begin{array}{c}-\frac{1}{(1-\mathrm{\&alpha;}){\mathrm{\&omega;}}_B},B_i\&le;B_{\min }\\ \frac{B_i-B_{\min }}{B_{\max }-B_{\min }},B_{\min }<B_i<B_{\max }\\ 1,B_i\&GreaterEqual;B_{\max }\end{array}\right.---\left(9\right)$

Standardization value for time delay

can be expressed as,

$f_{\mathrm{\&tau;}}^i=\left\{\begin{array}{c}-\frac{1}{(1-\mathrm{\&alpha;}){\mathrm{\&omega;}}_{\mathrm{\&tau;}}},{\mathrm{\&tau;}}_i\&GreaterEqual;{\mathrm{\&tau;}}_{\max }\\ \frac{{\mathrm{\&tau;}}_{\max }-{\mathrm{\&tau;}}_i}{{\mathrm{\&tau;}}_{\max }-{\mathrm{\&tau;}}_{\min }},{\mathrm{\&tau;}}_{\min }<{\mathrm{\&tau;}}_i<{\mathrm{\&tau;}}_{\max }\\ 1,{\mathrm{\&tau;}}_i\&le;{\mathrm{\&tau;}}_{\min }\end{array}\right.---\left(10\right)$

From formula 7 and formula 8, can find out, the standardization value of bandwidth and time delay all is not more than 1; When bandwidth/time delay does not meet the Minimum requirements of user's application, the standardization value substitution formula 7 by now, just can obtain the aggreggate utility value

minus; I.e. aggreggate utility value now very little, the possibility that is chosen as handover-target base station is very little; It is very good that this just can be avoided user terminal to be switched to most of performance parameter effectively, and go in the candidate base station that only has a parameter not meet consumers' demand;

Step 5, utilize step 3 and step 4, obtains respectively each base station BS in candidate network set B S _icorresponding aggreggate utility value

and the aggreggate utility value of Current Serving BTS if Current Serving BTS is not the optimal service base station, need to carry out handoff procedure;

$U_{\mathrm{total}}^{t\arg \mathrm{et}}=\max \left\{U_{\mathrm{total}}^i\right\}---\left(11\right).$

2. the adaptive vertical handoff method based on the residence time according to claim 1, is characterized in that user's residence time utilizes GPS information to calculate (being the GPS-MAV method), and method is as follows:

The position of WLAN base station is an O, and its coordinate is (x ₀, y ₀); The current location of terminal MN is a D, and its coordinate is (x _n, y _n); The present speed of terminal is big or small v _t, its direction and vector

identical; The definite ray of base station and terminal is

its middle conductor OD is the distance between base station and terminal; Ray identical with x axle direction; The terminal direction of motion

with x axle forward

angulation is

and

base station and terminal line

with x axle forward

angle be θ _t, and 0≤θ _t<2 π;

To terminal velocity v _tdo resolution of vectors; Its

the component size of direction is v _bS->MN, the component size of its vertical direction is v _vertical; Velocity component v _verticalnot affecting the distance between base station and terminal, is v apart from being subject to velocity component between base station and terminal _bS->MNimpact; In addition, v ' is another speed constantly of terminal, the angle corresponding with v '; According to top hypothesis, can express velocity component v _bS->MNand v _vertical,

If v _bS->MN0, the distance between terminal and base station increases, and received signal strength RSS reduces gradually; If v _bS->MN<0, the distance between terminal and base station reduces, and received signal strength RSS increases gradually; Angle theta _tit is vector

with vector

the angle become, be (x _n-x ₀, y _n-y ₀) angle definite with (1,0); So

${\mathrm{\&theta;}}_t=\left\{\begin{array}{c}\arccos \left(\frac{x_N-x_0}{\sqrt{{(x_N-x_0)}^2+{(y_N-y_0)}^2}}\right),(y_N-y_0)\&GreaterEqual;0\\ 2\mathrm{\&pi;}-\arccos \left(\frac{x_t-x_0}{\sqrt{{(x_N-x_0)}^2+{(y_N-y_0)}^2}}\right),(y_N-y_0)<0\end{array}\right.---\left(14\right)$

The end coordinates of previous moment is (x _n-1, y _n-1); Because the time interval that adjacent twice position upgraded is very little, can think that terminal does linear uniform motion within this cycle; Thereby the direction of motion of terminal just can be by the position coordinates (x in these two moment in one-period _n, y _n) and (x _n-1, y _n-1) determine,

Consider the randomness of user movement, carrying out speed while decomposing, employing be angle in nearest a period of time

mean value

In formula, α is index distribution smoothing factor,

by positional information p _kand p _k-1calculate;

The current distance d of terminal and WLAN base station _tfor

$d_t=\sqrt{{(x_N-x_0)}^2+{(y_N-y_0)}^2}---\left(17\right)$

The positional information p that terminal is up-to-date _nand p _n-1movement velocity size v _tfor,

$v_t=\frac{\left|\mathrm{\&Delta;d}\right|}{\mathrm{\&Delta;t}}=\frac{\sqrt{{(x_N-x_{N-1})}^2+{(y_N-y_{N-1})}^2}}{|t_N-t_{N-1}|}---\left(18\right)$

Consider the randomness of user movement, carrying out speed while decomposing, employing be velocity magnitude v in a period of time _tmean value

${\tilde{v}}_t=(1-\mathrm{\&alpha;})\underset{k=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{\&alpha;}}^{N-k}{v}_k---\left(19\right)$

In formula, v _kby positional information p _kand p _k-1calculate;

So speed exists

the component size v of direction _bS->NNcomponent size can be expressed as

If GPS updating location information cycle T _gPSvery little and can obtain continuously sample value, so, angle mean value

velocity magnitude mean value

just can be expressed as,

${\tilde{v}}_t=(1-\mathrm{\&alpha;}){\&Integral;}_0^N{\mathrm{\&alpha;}}^{N-i}{v}_i\mathrm{di}---\left(22\right)$

It is not constant that the randomness of user movement makes velocity magnitude and direction; A kind of situation is exactly to cover edge at certain point base stations to do back and forth movement; Now, the average speed size of terminal

value may be very large, but the displacement within a period of time is very little; If while according to above formula, asking the residence time, the result obtained has very large error, at this, proposes a new argument, i.e. effective displacement factor-beta,

$\mathrm{\&beta;}=\frac{|d_N-d_0|}{\underset{i=1}{\overset{N}{\mathrm{\&Sigma;}}}{\mathrm{dis}}_{i,i-1}}---\left(23\right)$

${\mathrm{dis}}_{i,i-1}=\sqrt{{(x_i-x_{i-1})}^2+{(y_i-y_{i-1})}^2}---\left(24\right)$

Effectively the displacement factor-beta has reacted the scale that terminal moves with respect to origin-location;

Estimate the residence time t of user in the current base station coverage _dwell,

$t_{\mathrm{dwell}\_\mathrm{dispose}}=\frac{R_{\mathrm{WLAN}}-d_t}{\left|v_{\mathrm{BS}->\mathrm{MN}}\right|}---\left(25\right)$

$t_{\mathrm{dwell}\_\mathrm{factor}}=\frac{R_{\mathrm{WLAN}}-d_t}{\mathrm{\&beta;}\left|{\tilde{v}}_t\right|}---\left(26\right)$

t _dwell=min{t _{dwell_dispose}，t _{dwell_factor}} (27)。

Images